Problems with installing SSL-certifates on ISPConfig3 slaveservers

Discussion in 'General' started by Hans, Oct 5, 2010.

  1. Hans

    Hans Moderator ISPConfig Developer

    Dear Till,
    Within these forums i can see that several persons do have problems with installing a SSL-certifcate for their website which are hosted on their ISPConfig3 server. Also do i and it causes me pain in my head.

    I try to explain what happens:
    I have a cluster of 6 ISPConfig3 servers (2 machines, 3 on each machine). These servers are all XEN VMs
    Every ISPConfig3 server (of course) has its own IP-address. On the master ISPConfig3 server i use some additional IP-addresses as well for the websites whith Comodo SSL-certifaces. This works excellent.
    However, when i define additional IP-addresses with version within ISPConfig3.0.2.2 it goes wrong.
    I mentioned this already before within thread
    After modifying the file /etc/network/interfaces again, i installed the website with a dedicated (additional) IP-address (not *) on the slave server.
    And yes i know which steps to take as i did this procedure many times.
    After installing the Comodo SSL-certicate everything is up and running, so thats good....alt least for a while.
    After some time Firefox shows me the error at http://www.example.tld "It works" and at https://www.example.tld ssl_error_rx_record_too_long. After some time everything is normal again, which means the site is up and running.

    First i thought: probably i did something wrong, maybe an DNS issue, but no. Also the vhosts where fine!
    I repeated all the steps 3 times for 2 websites with a SSL-certicate but the problem remains.

    After all, i decided to change the IP-addresses of the websites with SSL-certicates into the same IP-address of the host, which are the same IP-address of the slave ISPConfig3 servers.
    That worked and keeps working. The sites with SSL-certicates keeps up and running.

    On the master ISPConfig3 server i use websites with additional IP-addresses with SSL-certicates as well. No problems on the master server.
    However, the same result on the slave servers when additional IP-addresses is NOT possible.

    I want to ask you to have a look at this, because this is really not nice!
  2. till

    till Super Moderator Staff Member ISPConfig Developer

    This seems to be a DNS problem. Changing the IP address resolves such dns problems when the slave and master servers have diffenet IP addresses for a specific A-Record (split brain).

    You have to check the master and every slave server if they all give the correct IP address back for a specific A-record.

    ISPConfig is not doing any scheduled reconfigurations and if the vhost file is ok, then it is not caused by ispconfig.
  3. Hans

    Hans Moderator ISPConfig Developer

    Hi Till,
    Ok i see and thanks for your answer!

    To be honest, I did not think about a "split brain".
    But if so, how is that possible. I mean, all server domain names and websites make use of the same DNS-servers.
    These are the 3 DNS-servers of my DNS-registar. (I do not use my own.)
    Does that mean it is a DNS-cache problem on one of the DNS-servers i make use of?
  4. till

    till Super Moderator Staff Member ISPConfig Developer

    A split brain happens e.g. if one of the dns servers misses a dns update or did not accept the update. You can try to check the domain with a dns test tool like this:

    But it might be that it reports that everything is ok now as you changed the IP.
  5. Hans

    Hans Moderator ISPConfig Developer

    Hi Till,
    Thanks for your help and sorry for my wrong conclusion, but i was thinking that it must be a bug, because of my bad experience with adding additional IPs and the wrong result in /etc/network/interfaces.
    I did change the IP indeed and that avoids the problem exactly as you told me. :)

Share This Page